Plant Molecular Biology (2019) 99:477–497 https://doi.org/10.1007/s11103-019-00831-z

Opposite fates of the purine metabolite allantoin under water and nitrogen limitations in bread wheat

Alberto Casartelli1,6 · Vanessa J. Melino1,2 · Ute Baumann1 · Matteo Riboni1 · Radoslaw Suchecki1 · Nirupama S. Jayasinghe3 · Himasha Mendis3 · Mutsumi Watanabe4,7 · Alexander Erban4 · Ellen Zuther4 · Rainer Hoefgen4 · Ute Roessner3 · Mamoru Okamoto1 · Sigrid Heuer1,5

Received: 20 July 2018 / Accepted: 24 January 2019 / Published online: 5 February 2019 © The Author(s) 2019

Abstract Key message Degradation of nitrogen-rich purines is tightly and oppositely regulated under drought and low nitrogen supply in bread wheat. Allantoin is a key target metabolite for improving nitrogen homeostasis under stress. Abstract The metabolite allantoin is an intermediate of the catabolism of purines (components of nucleotides) and is known for its housekeeping role in nitrogen (N) recycling and also for its function in N transport and storage in nodulated legumes. Allantoin was also shown to differentially accumulate upon abiotic stress in a range of plant species but little is known about its role in cereals. To address this, purine catabolic pathway genes were identified in hexaploid bread wheat and their chromosomal location was experimentally validated. A comparative study of two Australian bread wheat genotypes revealed a highly significant increase of allantoin (up to 29-fold) under drought. In contrast, allantoin significantly decreased (up to 22-fold) in response to N deficiency. The observed changes were accompanied by transcriptional adjustment of key purine catabolic genes, suggesting that the recycling of purine-derived N is tightly regulated under stress. We propose opposite fates of allantoin in plants under stress: the accumulation of allantoin under drought circumvents its degradation to ammonium + + (NH­ 4 ) thereby preventing N losses. On the other hand, under N deficiency, increasing the ­NH4 liberated via allantoin catabolism contributes towards the maintenance of N homeostasis.

Keywords Allantoin · Drought · Nitrogen deficiency · Nutrient recycling · Purine catabolism · Triticum aestivum

Electronic supplementary material The online version of this article (https​://doi.org/10.1007/s1110​3-019-00831​-z) contains Introduction supplementary material, which is available to authorized users. Nitrogen (N) is a macronutrient required in large quantities * Sigrid Heuer [email protected] for plant development and growth with N deficiency causing chlorosis in older leaves and significant yield losses. Under 1 School of Agriculture Food and Wine, The University N deficiency and natural senescence plants translocate avail- of Adelaide, Urrbrae, SA 5064, Australia able N from source tissues to sink tissues, such as young 2 School of Agriculture and Food, The University leaves (Masclaux-Daubresse et al. 2010) and developing of Melbourne, Parkville, VIC 3010, Australia grains, accounting for 60–92% of total grain N (Barbottin 3 Metabolomics Australia, The University of Melbourne, et al. 2005). The glutamine synthetase–glutamate synthase Parkville, VIC 3052, Australia (GS-GOGAT) cycle plays an important role in this process 4 Max Plank Institute of Molecular Plant Physiology, since it recycles N liberated from the catabolism of N-rich 14476 Potsdam, Golm, Germany macromolecules, such as protein and nucleic acids, into low- 5 Rothamsted Research, Plant Science Department, Harpenden, molecular-weight organic compounds for long-distance N Hertfordshire AL5 2JQ, UK transport (Lea and Miflin2010 ). 6 Present Address: Strube Research GmbH & Co. KG, Purines are the most abundant N heterocyclic com- 38387 Söllingen, Germany pounds in nature and are found in nucleic acids (DNA, 7 Present Address: Graduate School of Biological Sciences, RNA) and many other cellular components, such as ATP, Nara Institute of Science and Technology, Ikoma, GTP or NADH (Werner and Witte 2011). Plants undergo the Nara 630‑0192, Japan

Vol.:(0123456789)1 3 478 Plant Molecular Biology (2019) 99:477–497 complete breakdown of the purine ring via a catabolic path- N deficiency, however, early reports showed that AtALN way enabling the recycling of both carbon (C) and N (Fig. was strongly up-regulated when Arabidopsis seedlings were S2). Overall, the oxidation of one molecule of xanthine to grown under N starvation (Yang and Han 2004). one molecule of glyoxylate liberates three molecules of ­CO2 Given the potentially important role of allantoin in N + and four molecules of ammonium (NH­ 4 ), which are likely metabolism and stress tolerance but the limited information to be reassimilated by the GS-GOGAT cycle into amino available in cereals, the aim of the present study was to char- acids. The pathway starts with the conversion of xanthine acterise the purine catabolic pathway in bread wheat. For to urate catalysed by xanthine dehydrogenase (XDH) (Tri- this, we selected two wheat genotypes (RAC875 and Mace) plett et al. 1982; Werner and Witte 2011). Urate is further that are adapted to Australian environments. The gene loci processed by urate oxidase (UOX) producing 5-hydroxyi- were identified based on the reference genome of the cultivar sourate (5-HIU), 5-HIU is then converted to allantoin via the Chinese Spring and their chromosomal location was experi- 2-oxo-4-hydroxy-4-carboxy-5-ureido-imidazoline (OHCU) mentally verified. Quantification of allantoin in different tis- intermediate by allantoin synthase (AS) (Hanks et al. 1981; sues throughout plant development revealed accumulation of Ramazzina et al. 2006; Kim et al. 2007; Lamberto et al. allantoin under drought and reduced allantoin levels under 2010; Pessoa et al. 2010). Allantoin is metabolised to allan- N limitation and this was associated with differential regula- toate by allantoinase (ALN) and then to ureidoglycine by tion of the genes in the purine catabolic pathway. Monitor- allantoate amidohydrolase (AAH) (Yang and Han 2004; ing of N-metabolite pools in seeds revealed that allantoin Todd and Polacco 2006; Werner et al. 2008). The last two levels progressively increase in developing grains and that enzymatic steps are catalysed by ureidoglycine aminohy- genotypic differences exist for accumulation of allantoin and drolase (UGAH), which converts ureidoglycine to ureido- other important N-containing metabolites. glycolate (Serventi et al. 2010). Ureidoglycolate amidohy- drolase (UAH) converts ureidoglycolate to hydroxyglycine and, lastly, hydroxyglycine decays to glyoxylate by a non- Materials and methods enzymatic reaction (Werner et al. 2010). In addition to its housekeeping role in N recycling, the Plant material purine catabolic pathway has an important function in cer- tain dinitrogen ­(N2)-fixing legumes (Schubert 1986; Sinclair Two semi-dwarf South Australian wheat genotypes RAC875 and Serraj 1995; Alamillo et al. 2010; Coleto et al. 2014). and Mace were evaluated in this study. RAC875 (RAC655/3/ Allantoin and allantoate (also known as ureides) are the Sr21/4*LANCE//4*BAYONET) is high-yielding in the main products of atmospheric N­ 2 fixation in root nodules, drought and heat-prone South Australian environments which are then translocated to the shoot (Herridge et al. (Izanloo et al. 2008; Bennett et al. 2012). Mace (WYALK- 1978; Pate et al. 1980). In recent years, the ureide allantoin ATCHEM/STYLET//WYALKATCHEM) was bred and has gained attention by the scientific community as several released by Australian Grain Technologies (AGT) in 2008 metabolomics studies reported this metabolite to accumulate and preliminary studies suggest that Mace has high N-use in a broad range of plant species under drought (Bowne et al. efficiency across different South Australian environments 2011; Oliver et al. 2011; Silvente et al. 2012; Degenkolbe (Mahjourimajd et al. 2016). et al. 2013; Yobi et al. 2013; Casartelli et al. 2018), high salt (Kanani et al. 2010; Wu et al. 2012; Nam et al. 2015; Wang Plant growth et al. 2016), cold (Kaplan et al. 2004) and sulfate starvation (Nikiforova et al. 2005). In contrast, allantoin was found The experiment was conducted in a controlled environ- to be reduced under prolonged N deficiency in maize and ment with day/night cycle of 12 h/12 h at a flux density at rice (Amiour et al. 2012; Coneva et al. 2014). Further, the canopy level of 300 µmol m−2 s−1, 20 °C/15 °C day/night work of Watanabe et al. (2014) revealed that Arabidopsis temperature and 82% average humidity. Potting mixture AtALN mutants, which constitutively accumulated allantoin, was composed of river sand and coco-peat and prepared were more tolerant to desiccation stress. It was demonstrated according to Melino et al. (2015). Granular urea was pro- that allantoin mediates abscisic acid (ABA) signalling by vided as basal N application with rates of 150 and 75 mg N stimulating the activity of genes and enzymes belonging to kg­ −1 for high and low N treatments, respectively. At stem the ABA-producing pathways. In addition, allantoin was elongation (39 days after sowing) a third of the basal urea shown to accumulate in Arabidopsis leaves grown under rates for each treatment were applied by soil drenching. drought and salt stress in coordination with transcriptional A soil water retention curve was constructed by measur- changes of a number of purine catabolic genes (Irani and ing the pre-dawn leaf water potential of 3-week old seed- Todd 2016; Lescano et al. 2016). To date, there are few stud- lings under progressive drought stress with a Scholander- ies on the regulation of the purine catabolic pathway under type pressure chamber (Soil Moisture Equipment Corp.,

1 3 Plant Molecular Biology (2019) 99:477–497 479

Santa Barbara, USA) (Fig. S1b). Overall, the experiment additionally included in the gene name to allow distinction was comprised of two cultivars (RAC875 and Mace), of the three homeologous sequences (Table S1). three treatments (high nitrogen well-watered (HN-WW), low nitrogen well-watered (LN-WW) and high nitrogen drought (HN-D)), six sampling time points and six bio- Synteny analysis logical replicates for a total of 216 experimental units. Individual seeds were germinated in black square plastic Synteny analysis were conducted based on available pots (11 × 11 cm2 area, 14 cm height) containing 1.5 kg of genome sequences of Sorghum bicolor, Oryza sativa soil mix arranged in a randomised complete block design. and Brachypodium distachyon using the Phytozome 11 WW conditions were maintained by daily watering of ‘Ancestry’ tool (phytozome.jgi.doe.gov). Generally, five the pots to 20% soil water content (SWC), calculated as genes upstream and downstream of the respective purine SWC = [(mwet soil − mdry soil):mdry soil] × 100. Drought was catabolic pathway gene were included in the orthology induced by withholding water until signs of leaf rolling analysis. These genes in bread wheat were identified by appeared (approximately 6.5% SWC; Fig. S1a). Samples TBLASTN searches of Brachypodium genes against the collected during the reproductive stages were harvested TGACv1 wheat genome assembly (http://pre.plants.ensem​ ​ according to the anthesis date of individual plants and bl.org/Triticum_aesti​ vum/Info/Index​ ).​ Genes were consid- water was first withheld 2 weeks after anthesis. Whole ered syntenic if they were present on the same TGACv1 spikes were collected from the main stem of each plant scaffold where the wheat purine catabolic genes were and to minimise the differences in development of the annotated. grains along the spike, only the middle part of the spike was employed for further analyses. Rachis was removed and the vegetative part of the spike (hereafter referred as Wheat nulli‑tetrasomic lines spikelet) was separated from the developing grains. Details on the sample collection throughout the experiment are To verify the chromosomal location of the wheat purine given in Fig. S1a. Samples for molecular analysis were catabolic gene Chinese Spring nulli-tetrasomic (NT) lines snap-frozen in liquid nitrogen and stored at − 80 °C until were used in which individual wheat chromosomes were further use. replaced (nullisomic) by an extra pair (tetrasomic) of their homeologs (Sears 1954). PCR was performed using 150 ng of genomic DNA of the NT lines as template using standard Identification of the purine catabolic genes in wheat reaction conditions of ThermoPol (BioLabs, USA) using the and other grass genomes primers provided in Table S1.

The purine catabolic genes of Oryza sativa, Zea mays, Sorghum bicolor and Brachypodium distachyon were Allantoin quantification identified with a BLASTP search using Arabidopsis thali- ana protein sequences as a query in Phytozome (https​:// Metabolite extraction phyto​zome.jgi.doe.gov). B. distachyon protein sequences were then used for a TBLASTN search on the barley WGS Metabolites were extracted from 10 mg of homogenised, Morex Assembly version 3 (https://ics.hutto​ n.ac.uk/morex​ ​ freeze-dried tissue with 500 µl of 100% (v/v) methanol Genes​/blast​_page.html) and the barley predicted protein containing 12.5 µM 13C15N-allantoin (internal standard), sequences were used for a TBLASTN search against except for mature grains samples for which 25 µM 13C15N- the Chinese Spring TGACv1 genome assembly (http:// allantoin was used. Samples were vortexed and incubated pre.plant​s.ensem​bl.org/Triti​cum_aesti​vum/Info/Index​) in an Eppendorf Thermomixer at 1400 rpm and 30 °C for (Clavijo et al. 2017). In cases where full- length sequences 15 min followed by a 15 min centrifugation at 13,000 rpm were not found in Chinese Spring TGACv1, we searched (4 °C). The supernatant was transferred to a new tube. 500 µl the Chinese Spring IWGSC chromosome survey sequence of milli-Q water was added to the remaining pellet, vortexed (css) version 3 (IWGSC 2014—http://www.wheat​genom​ and centrifuged for 15 min at 15,000 rpm. The supernatant e.org) and the sequences were merged with TGACv1 was combined with the previous one, vortexed for 30 s and sequences using Geneious version 10.0.2 (http://www. centrifuged for 15 min at 15,000 rpm. The resulting superna- geneious.com​ ) (see Table S1 for details). Each purine cata- tant was transferred to a new tube and 300 µl of 100% (v/v) bolic gene was assigned with the name of the Arabidop- chloroform was added, vortexed and centrifuged for 5 min at sis orthologous according to Watanabe et al. (2014). For 15,000 rpm. 800 µl of the top (polar) phase was transferred wheat, the sub-genome localisation (e.g., Chr 1AL) was into a new tube for allantoin analysis.

1 3 480 Plant Molecular Biology (2019) 99:477–497

Sample derivatization detection for the allantoin MRM, was 0.294 µM based on a signal to noise ratio of 3. 200 µl of the upper polar phase was aliquoted in a glass Data was processed using Agilent MassHunter QQQ insert and dried under vacuum (RVC 2-33 CD plus, John Quantitative Analysis software (B.07.00). Allantoin was Morris Scientific Australia) set at ambient temperature. quantified by single point calibration based on the relative All samples were re-constituted in 10 µl of methoxy- response (the response area of allantoin (MRM 398 → 171) amine hydrochloride (30 mg ­ml−1 pyridine) and deriva- divided by the response area of 13C15N-allantoin (MRM tised at 45 °C for 60 min at 500 rpm before adding 20 µl 400 → 173)) and 13C15N-allantoin concentration. Sample dry of N-methyl-N-(tert-butyldimethylsilyl)trifluoroaceta- weight and extraction volume were taken into consideration mide (MTBSTFA) with 1% (w/v) trimethyl chlorosilane when calculating the final allantoin concentration. (TMCS) and incubated at 45 °C for 45 min.

Metabolome‑wide analysis GC–MS instrument conditions Metabolites were extracted from 60 mg dry weight (DW) 1 µl of derivatised sample was injected into a GC-QqQ- of freeze-dried wheat spikelet and developing grain sam- MS system comprised of a Gerstel 2.5.2 Autosampler, a ples using 100% methanol containing 13C-sorbitol as inter- 7890A Agilent gas chromatograph and a 7000 Agilent nal standard. Chloroform followed by water were added to triple-quadrupole MS (Agilent, Santa Clara, USA) with the mixture. Polar phase aliquots were freeze-dried using a an electron impact (EI) ion source. The GC was operated speed vacuum. Aliquots for GC-time of flight (TOF)-MS in constant flow mode with helium as carrier gas. The MS analysis were reconstituted in N-Methyl-N-(trimethylsilyl) was adjusted according to the manufacturer’s recommen- trifluoroacetamide (MSTFA) and methoxyamine hydro- dations using tris-(perfluorobutyl)-amine (CF43). A J&W chloride as derivatizing agents. Samples were analysed as Scientific VF-5MS column (30 m long with 10 m guard described previously by Lisec et al. (2006) and Erban et al. column, 0.25 mm inner diameter, 0.25 µm film thickness) (2007). Aliquots for ion chromatography (IC) were recon- was used. The injection temperature was set at 250 °C, the stituted in milli-Q water and analysed according to Moschen MS transfer line at 290 °C, the ion source was adjusted to et al. (2016). Aliquots for free spermidine analysis were 230 °C and the quadrupole at 150 °C. Helium was used reconstituted in 0.2 N perchloric acid solution, dansylated as carrier gas at flow rate of 1 mlmin ­ −1. Nitrogen (UHP and quantified by HPLC according to Do et al. (2013). 5.0) was used as collision cell gas at flow rate of 1.5 ml ­min−1. Helium (UHP 5.0) was used as quenching gas at a flow rate of 2.25 ml ­min−1. Gain factor for the triple axis Amino acid analysis detector was set at 2. Derivatised sample was injected into the column at 100 °C followed by 1 min hold, followed by Amino acids were extracted from 50 mg DW of freeze-dried a ramp of 25 °C ­min−1 to 325 °C. wheat samples with 1 ml of 10 mM sodium acetate contain- ing 250 nmol ml−1 Norvaline (internal standard). Amino acids were quantified on a Waters Acquity™ UPLC system Method optimization using the Waters AccQ-Tag Ultra Chemistry Kit follow- ing the manufacturer’s instructions (Waters Corp., USA). Allantoin and 13C15N-allantoin standards were purchased Chromatograms were analysed with Empower® 3 Software from Sigma Aldrich (Australia). Retention time, a corre- (Waters Corp., USA). sponding unique precursor ion and product ions were iden- tified on the GC-QqQ-MS instrument for each standard. Collision energy was optimised between 0 and 20 V for Total N analysis the identified precursor to product ion transitions (Multiple Reaction Monitoring; MRM). A product ion was selected as Total N concentration was determined using the Elementar the Target ion (T) and the other subsequent MRM transition rapid N ­exceed® (Elementar Analysensysteme GmbHana- was set as the Qualifier ion (Q) (see Table S2 for details). lyser, Germany) using 100 mg DW of homogenised wheat Linearity of the method was tested using serial dilutions grain samples. Aspartic acid (250 mg) was used as standard of the calibration standard and showed a linear calibration for calibration: the theoretical aspartic acid N% (10.52) was range between 0.98 µM to 250 µM 13C15N-allantoin. Cor- divided by the N% measured by the instrument generating relation coefficient ­(R2) of the calibration curve was 0.99 a N factor. The N factor was then used to correct the N% for the target allantoin MRM (398 → 171). The limit of measured for each sample.

1 3 Plant Molecular Biology (2019) 99:477–497 481

RNA‑Seq reads mapping for the TaXDH2 paralogous manufacturers’ instructions. cDNA quality was verified by genes PCR amplification of the actin gene (Table S3).

Publicly available bread wheat (cv. Chinese Spring) RNA- Primer design and quantitative reverse Seq data was used for generating the per-base expression transcription PCR (qRT‑PCR) analysis values. This data set covers 15 duplicated samples corre- sponding to five organs (root, leaf, stem, spike, grain), each Quantitative real-time PCR was performed with KAPA at three developmental stages (http://urgi.versa​illes​.inra. ® fr/files​/RNASe​qWhea​t/) (Choulet et al. 2014). The RNA- SYBR­ Fast qPCR kit Master Mix, and amplification was Seq reads were quality-, adapter- and length-trimmed using real-time monitored on a QuantStudio™ 6 Flex Real-Time Trimmomatic (Bolger et al. 2014), version 0.30 with a cus- PCR System (Applied Biosystems, USA). Gene-specific tom list of adapter sequences and the following settings: primers targeted to amplify all three homeologs simulta- ‘ILLUMINACLIP:adapters.fa:1:6:6 LEADING:3 TRAIL- neously were designed with AlleleID® software (Premiere ING:3 SLIDINGWINDOW:4:6 MINLEN:60’. The reads Biosoft) (Table S3) and the specificity of each pair was were aligned to the scaffolds (version 3) from the Chinese verified by melting curve analysis and sequencing of the Spring whole genome assembly (version 0.4) using STAR products. Change in gene expression were calculated using (version 2.5.1b) (Dobin et al. 2013), with the following set- qBASE + software and reported as calibrated normalised tings: --outFilterMultimapScoreRange 0; --outFilterMul- relative quantities (CNRQ) that represents the relative gene timapNmax 5; --outFilterMismatchNoverLmax 0 --out- expression level between different samples for a given target FilterMatchNminOverLread 1; --outSJfilterOverhangMin gene: 35,20,20,20; --outSJfilterCountTotalMin 10,3,3,3; --outSJfil- terCountUniqueMin 5,1,1,1 --alignEndsType EndToEnd; --alignSoftClipAtReferenceEnds No; --alignIntronMax 10,000; --alignMatesGapMax 10,000. The remaining set- tings were left at their defaults. The resulting BAM files were merged using samtools merge (version 1.2) (Li et al. 2009). The manually annotated coordinates of the genes define the regions for which the depth of the aligned reads NRQ is then divided by a calibration factor (CF) (Hel- was computed from the aligned BAM file using the depth lemans et al. 2007). Four reference genes (Table S3) were module of samtools. quantified by qRT-PCR: TaActin, TaGAPdH, TaCyclo- philin and TaEFA2. CNRQ values were calculated using RNA preparation the most stable genes within a specific tissue (selected by qBASE + software). cDNA libraries were prepared from snap-frozen samples of the youngest fully emerged leaf (YFEL), flag leaf, stem, spikelet and developing grain under control (HN-WW) Data analysis and treatment (HN-D and LN-WW) conditions. Samples were ground to a fine powder using a 2010 Geno/Grinder® Statistical analyses were performed using GraphPad Prism (SPEX SamplePrep, USA) and total RNA was extracted version 7.00 for Windows (GraphPad Software, La Jolla from frozen samples with a phenol and guanidine thiocy- California USA. http://www.graph​pad.com). All data are anate buffer according to Chomczynski (1993). To extract reported as mean ± SEM. Significant differences between RNA from developing grains which have a high polysac- means of two groups of data were tested by Student’s t-test. charide content, an extraction buffer (1% (w/v) sarcosyl, Significant differences between means of more than two 150 mM NaCl, pH 9) and a guanidine hydrochloride-based groups were tested by two-way ANOVA. For metabolome- buffer for purification according to Singh et al. (2003) was wide analysis, metabolites levels were log transformed to employed. Genomic DNA was removed using the TURBO improve the normality of the data set and then scaled by DNA-free™ Kit (Ambion®, Thermo Fisher Scientific, USA) subtracting the median metabolite value in each metabo- following the manufacturers’ instructions. High RNA quality lite distribution. Hierarchical clustering by Pearson’s cor- was confirmed in a 2% (w/v) agarose gel visualized under relation distance and PCA analyses were performed with UV light and RNA concentrations were quantified using the support of ClustVis web tool (Metsalu and Vilo 2015). a ND-1000 spectrophotometer (NanoDrop Technologies, Transcriptional data was reported as log2 ratio of the fold- USA). 1 µg of RNA was used for cDNA synthesis using the change between treatment (drought or low N) and control SuperScript® III kit (Thermo Fisher Scientific, USA) as per conditions.

1 3 482 Plant Molecular Biology (2019) 99:477–497

Fig. 1 Distribution of the purine catabolic genes across the bread wheat allohexaploid genome (Triticum aestivum cv. Chinese Spring). a The 24 purine catabolic genes are represented by grey bars and their putative location on chromosomes was estimated based on the related Munich Information Centre for Protein Sequences (MIPS) gene annotation on Ensem- blPlants (plants.ensembl.org). The figure was adapted from Gill et al. (2004). b Homoelog- specific primers were used for PCR amplification of the genes using DNA derived from nulli-tetrasomic lines of the cv. Chinese Spring. Labels shown in the agarose gel indicate the nullisomic chromosome, e.g. “N1A” indicates that chro- mosomal group 1A is absent (see Table S1 for details). The caption indicates the primer set employed to amplify the NT DNA (details provided in Table S1). For example, (i) XDH1-1AL,-1BL and − 1DL primer sets were used to amplify nullisomic lines for chromosome subgroup 1. The absence of an amplicon in a specific NT line (whilst ampli- cons are derived in the two remaining NT lines) indicates localisation of the gene on the respective absent chromosome

Results short arm of chromosome group 6 (TaXDH2-6AS/TaXDH2- 6BS/TaXDH2-6DS), respectively. The TaXDH1 and TaXDH2 Identification of the wheat purine catabolic genes paralogs share 89.6–90.5% sequence identity within the predicted coding region (CDS) and the TaXDH1 home- Based on comparative sequence analyses, a total of 24 wheat ologs showed a higher CDS identity (98.6–98.8%) than the genes that are homologous to the purine catabolic genes of TaXDH2 homeologs (93.5–95.3%) (Table S4). The TaALN Arabidopsis and rice were identified (Fig. 1a). Generally, genes (TaALN-2AL/TaALN-2BL/TaALN-2DL) and TaU- for each rice gene three wheat homeologous sequences GAH genes (TaUGAH-2AS/TaUGAH-2BS/TaUGAH-2DS) were identified, except for XDH that had paralogous copies were localised on chromosome group 2, whilst the TaUOX on two different chromosome groups. Specifically, the two genes were localised on chromosome group 3 (TaUOX- allelic variants were located on the long arm of chromosome 3AL/TaUOX-3BL/TaUOX-3DL). The TaAS genes were iden- group 1 (TaXDH1-1AL/TaXDH1-1BL/TaXDH1-1DL) and tified on chromosome group 4 and theTaUAH genes were

1 3 Plant Molecular Biology (2019) 99:477–497 483 localised on chromosome group 5 (TaUAH-5AS/TaUAH- with either high N (HN) or low N (LN). During the course 5BS/TaUAH-5DS). The TaAAH genes were found located of the experiment, a subset of plants grown under HN were on chromosome group 7 (TaAAH-7AL/TaAAH-7BL/TaAAH- subjected to drought (D) at tillering stage and during grain 7DL) and were the least conserved among the analysed filling, (Fig. S1a). Therefore, the experiment comprised of genes with 95–92.9% sequence identity within the predicted three treatments: control (high N, well-watered; HN-WW), CDS (Table S4). low N (low N, well-watered; LN-WW) and drought (high To experimentally validate the chromosomal localization N, drought; HN-D). Analysis of plants at maturity revealed of the wheat orthologues genes predicted by the TAGCv1 that the low N treatment reduced above ground biomass by and IWGSC css assembly, nulli-tetrasomic (NT) lines of the 20.8% and 23.3% in RAC875 and Mace, respectively, whilst wheat cultivar Chinese Spring were used (Fig. 1b; Sears grain yield was reduced by 24.4% and 25.2% in RAC875 and 1954; see M&M for details) and homeolog-specific primers Mace as compared to the controls (HN-WW), respectively (Table S1) designed for PCR analysis of genomic DNA from (Fig. 3). When plants were grown under drought conditions, the NT lines. The absence of an amplicon in the respective above-ground biomass was reduced by approximately 15% NT line confirms that the target gene is physically located in both genotypes as compared to their respective controls. on that nullisomic chromosome pair. Under drought conditions, Mace plants were also higher To further corroborate the orthology between the identi- yielding (grain yield) than RAC875 plants (p < 0.02 by Stu- fied loci we analysed the synteny of the genomic regions dent’s t-test); this corresponded to a 16.0% and 26.6% yield between bread wheat, Brachypodium, rice and sorghum loss for Mace and RAC875, respectively, relative to control (for gene IDs see Supplementary Tables S1 and S5). The conditions (Fig. 3). fragmented nature of the bread wheat TGACv1 assembly (Clavijo et al. 2017) may have reduced the resolution of the Allantoin accumulation under low N and drought analysis, however, wheat showed a high degree of synteny stress with the three diploid genomes included in the analysis and the position of several purine catabolic genes and their To assess whether N and water stress alter allantoin levels neighbouring genes was highly conserved (Fig. 2). The only in the two wheat genotypes, allantoin concentration was exception was XDH, for which we could identify only one quantified in the youngest fully emerged leaf (YFEL) during syntenic gene (Bradi1g15910, depicted in blue) among all vegetative growth and in flag leaf, stem, spikelet and devel- analysed genomes. This gene is located upstream of XDH in oping grain samples during reproductive growth in plants sorghum, downstream of XDH in rice and Brachypodium, grown under HN-WW, LN-WW and HN-D (Fig. 4). Nitro- and on chromosome group 1 in wheat. Interestingly, we gen deficiency significantly reduced allantoin concentration identified two Brachypodium genes (Bradi1g15820 and Bra- at different time points across different tissues and in both di1g15826, depicted in brown and grey, respectively) with genotypes when compared to control conditions (Fig. 4a). In orthologous sequences on wheat chromosome group 6, sup- both genotypes grown under LN-WW conditions, allantoin porting the evidence that a duplication of the genomic region levels were generally below 50 nmol ­g−1DW in YFEL, flag harbouring XDH may have occurred during wheat genome leaves and stems, but much higher (50 to 100 nmol g­ −1DW) evolution. Analysis of the predicted open reading frames in spikelets and developing grains. The highest reduction of (ORFs) of the TaXDH genes revealed five premature stop allantoin in both genotypes was measured in the stem under codons in TaXDH2-6AS and several mutations in TaXDH2- LN-WW as compared to the HN-WW control, with a 22-fold 6BS, both likely to result in non-functional proteins. In con- and 10-fold reduction in RAC875 and Mace, respectively trast, the TaXDH2-6DS ORF appeared to translate into a (Fig. 4a). Interestingly, under control conditions (HN-WW) functional protein (data not shown). This was supported by Mace flag leaves significantly accumulated much higher in-silico analysis of publicly available Chinese Spring RNA- levels of allantoin at 19 DAA (305 nmol ­g−1 DW) as com- seq transcriptomics data which revealed that TaXDH2-6DS pared to RAC875 (34 nmol ­g−1 DW), whilst a significant was the only homeolog expressed (Fig. S3a). Synteny analy- but relatively smaller difference was recorded under low N sis further revealed the presence of two copies of UAH in the at 17 DAA (36 and 12 nmol g­ −1 DW for Mace and RAC875, Brachypodium genome of which only one gene (BdUAH1) respectively) (Fig. 4a). Similarly, allantoin concentration in showed a syntenic relationship with the other genomes. Mace developing grains was significantly higher compared to RAC875 at 22DAA under both control (HN-WW) and low The effects of low N and drought on plant N (LN-WW) conditions. performance In contrast to the reduced accumulation of allantoin under N deficiency, allantoin significantly positively accu- RAC875 and Mace plants were initially grown under well- mulated under drought in all tissues assessed, with the watered (WW) conditions divided in two subsets supplied exception of the stem where, despite a clear positive trend,

1 3 484 Plant Molecular Biology (2019) 99:477–497 H H H H AA AA AA AA N N N AA H AA H N N N AL AL AL AL AL AL BL AL 7DL r7 Ch r7 Ch Chr 0 DL BL AL r6 Ch r1 r2 r2 r2 Ch Ch r4 Ch Ch Ch r6 Ch r9 Ch r5 ALNS ALNS ALNS ALNS ALNS ALNS Ch BL DL AS r4 r4 r4 Ch Ch Ch X X X X X X UO UO UO UO UO UO AL BL Ch r3 Ch r3 DL Ch r3 r2 Ch XDH1 XDH1 XDH1 BL AL DL r1 r3 r1 r1 r1 Ch Ch Ch Ch Ch H XD H H XD XD r3 r1 Ch Ch XDH2 XDH2 XDH2 AS DS 6BS r6 r6 r1 AL Chr Ch Ch Ch r2 Ch H2 UGAH UGAH UGAH UGAH UGAH UA UGAH r2 BS AS DS H H H H H1 H Ch r2 r2 r2 UA UA UA UA UA UA Bd Ch Ch Ch Ch r7 r1 Ch BS AS 5DS r5 r5 r2 Ch Ch Chr Ch r4 Ch 2 r1 Ch m heat heat podium podi Syntenic relations of the genomic regions harbouring purine catabolic genes in sorghum, rice, Brachypodium and wheat. The four species are shown in four distinct rows underneath each rows distinct four in shown are species The four wheat. and harbouringregions the of relations Syntenic rice, genomic purine sorghum, in Brachypodium catabolic genes r8 dw dw hy hy Ch orghum orghum Brea Brac Rice S Brea S Brac Rice 2 Fig. the of the presence shape in different same coloured gene, purine each indicated. For of the catabolic respective genes shapes with as coloured represented are abbreviations other and genes genome distances to scale and therefore drawn is not number is indicated. This figure orthologs. and the the chromosome of gene denotes presence chromosomes Solid lines represent genomes the(GOI). In genome, wheat bread interest of the adjacent to genes but not determinedbe the on the located cannot are from length same chromosome indicate thatlines Dashed lines. of genes as scaffold but assembled on purine scaffold the on a different identified were on represented are genes the same solid line. Genes catabolicthat on the located same TGACv1 that were genes a dashed line the from separated GOI by are the same chromosome

1 3 Plant Molecular Biology (2019) 99:477–497 485

(A)(Biomass B) Grain yield RAC875 15 5 aa * Mace 1 - aa t 4 b b

1 b an b 10 bb bb - t

pl 3 n ) la

p 2 DW 5 g g( 1

0 0 HN-WWLN-WW HN-D HN-WWLN-WW HN-D

Fig. 3 Analysis of agronomic traits of plants at maturity. a Above and letters indicate significant differences between genotypes and ground biomass (g DW ­plant−1) and b grain yield (g ­plant−1) were treatments at p < 0.05. The analysis revealed a significant treatment measured from the remaining tillers of wheat RAC875 and Mace effect (p < 0.0001) for both traits and a significant genotype × treat- plants used for molecular analyses. Plants were grown under control ment interaction (p < 0.05) for grain yield. Asterisk denotes signifi- (HN-WW), low N (LN-WW) and drought (HN-D) conditions and cant difference (p < 0.02) between grain yield of RAC875 and Mace each value represent the mean ± SEM of 16–18 biological replicates. under drought according to Student’s t-test Two-way ANOVA analysis was performed with Tukey’s correction

(A)

(B)

Fig. 4 Allantoin concentration in bread wheat genotypes RAC875 corresponding to 3, 5 and 8 days of drought (days after last water- and Mace under N deficiency and drought stress. Allantoin concen- ing), respectively (see Fig. S1 for details). Data are mean ± SEM of tration (nmol g­ −1DW) was quantified by GC-QqQ-MS undera low N four to six biological replicates and letters indicate significant differ- and b drought conditions with respect to control conditions in young- ences between treatments and time points within a genotype (lower est fully emerged leaf (YFEL), flag leaf, stem, spikelet and develop- case: RAC875; upper case: Mace) by two-way ANOVA with Tukey’s ing grain. YFEL samples were collected during tillering stages at correction at p < 0.05. Asterisks indicate significant allantoin differ- 23, 24 and 27 days after sowing, corresponding to 2, 3 and 6 days ences between the genotypes RAC875 and Mace for a given treat- of drought stress (days after last watering), respectively. Flag leaf, ment and time point by Student’s t-test, corrected for multiple com- stem, spikelet and developing grain samples were collected dur- parison using the Bonferroni-Dunn method (*p < 0.05; **p < 0.01; ing grain filling stages at 17, 19 and 22 days after anthesis (DAA), ***p < 0.001; ****p < 0.0001 by Student’s t-test)

1 3 486 Plant Molecular Biology (2019) 99:477–497 allantoin accumulation was not significant (Fig. 4b). Dur- most severe (Fig. 4b). Interestingly, flag leaves of RAC875 ing the tillering stage, under severe drought stress (day 8; progressively accumulated allantoin during the course of 6.3% SWC) allantoin significantly accumulated 3.6- and the drought treatment, increasing from 136 nmol ­g−1 DW 2.5-fold in the YFEL of RAC875 and Mace, respectively. at day 5 to 1197 nmol g­ −1 DW at day 8, representing an The higher magnitude of allantoin accumulation under almost 30-fold increase relative to HN-WW. On the other drought in RAC875 were due to significantly higher levels hand, Mace accumulated 1614 nmol of allantoin per gram of allantoin in the YFEL of RAC875 under drought com- DW after only 5 days of drought treatment and levels pared to Mace (Fig. 4b). Similarly, during grain filling, remained constant until day 8. However, the magnitude of allantoin significantly accumulated in Mace flag leaves allantoin accumulation under HN-D in Mace was only 2.5- at day 5 and 8 of the drought treatment (8.6% and 6.6% fold as compared to HN-WW at day 8 (Fig. 4b). The large SWC, respectively), whilst it significantly accumulated in difference in the magnitude of allantoin accumulation RAC875 flag leaves only at day 8 (Fig. 4b). In spikelet between RAC875 and Mace flag leaf can be associated and developing grains, allantoin significantly increased with the higher allantoin levels in Mace already present only at day 8 in both genotypes, when drought stress was under control conditions at day 5 (19DAA).

YFEL Flag leaf Stem Spikelet Developing grain LowN (Z22) (Z79) (Z79) (Z79) (Z79) Log2 CNRQ (LN/HN) 23 DAS24DAS 27 DAS17DAA 19 DAA22DAA 17 DAA19DAA 22 DAA17DAA 19 DAA22DAA 17 DAA19DAA 22 DAA

TaXDH1 0.12 0.33 0.38 0.94 2.03 -0.050.420.400.540.010.290.090.52-0.05 0.79 TaXDH2 -0.07-0.08 0.40 -0.491.390.860.190.570.450.060.400.211.940.170.81 TaUOX 0.22 0.27 0.26 0.52 1.94 0.44 0.08 0.44 0.28 0.09 0.18 0.02 0.63 -0.170.12 TaALNS 0.11 0.28 0.05 0.82 1.65 0.45 -0.080.300.240.010.17-0.07 0.30 0.39 0.33 TaALN 1.78 1.00 1.12 0.59 1.42 1.14 1.91 1.16 0.45 1.33 0.82 0.60 0.60 0.26 0.25 TaAAH 0.60 1.90 0.85 1.50 2.37 2.74 1.24 1.08 1.39 0.79 0.54 0.71 0.50 -0.010.15 TaUGAH -0.02-0.09 -0.150.210.610.020.090.480.32-0.09 0.22 0.06 0.84 0.26 -0.25 TaUAH -0.030.160.090.271.720.450.360.260.31-0.08 0.30 -0.140.44-0.11 0.12

YFEL Flag leaf Stem Spikelet Developing grain Drought (Z22) (Z79) (Z79) (Z79) (Z79) Log2 CNRQ (D/WW) Day2 Day3 Day6 Day3 Day5 Day8 Day3 Day5 Day8 Day3 Day5 Day8 Day3 Day5 Day8

TaXDH1 0.16 0.46 0.63 -0.320.362.620.01-0.06 0.89 -0.110.922.060.06-0.14 0.14 TaXDH2 -0.44-0.90 -0.22-1.48 -1.04-0.74 -0.23-0.29 -1.37-0.52 -0.14-0.14 0.96 0.06 0.09 TaUOX 0.16 0.19 0.66 -0.220.781.06-0.12 -0.060.14-0.14 0.81 1.49 0.31 -0.27-0.54 TaALNS 0.20 0.41 0.11 -0.070.53-1.04 -0.14-0.13 -0.45-0.01 0.51 0.50 -0.040.16-0.85 TaALN -0.78-1.57 -1.53-1.18 -1.88-0.07 0.60 -1.10-2.07 0.49 -0.26-0.09 0.03 -0.79-2.86 TaAAH -0.020.800.210.08-0.49 1.38 0.14 -0.55-1.21 0.01 0.20 0.43 0.16 -0.64-1.19 TaUGAH -0.30-0.14 0.31 -0.44-0.64 -2.89-0.19 -0.44-1.36 -0.210.42-0.08 0.66 0.19 -1.01 TaUAH -0.120.380.85-0.37 0.24 1.71 0.08 0.07 0.58 -0.060.861.70-0.01 -0.27-0.72

log2 -3 +3

Fig. 5 qRT-PCR analysis of the purine catabolic genes in different grain. Data is expressed as log2 calibrated normalised relative quan- tissues of the bread wheat genotype RAC875 under stress. RNA was tity (CNRQ) of gene transcription under N deficiency or drought extracted from the youngest fully emerged leaf (YFEL), flag leaf, divided by CNRQ under control. Colour denotes significant differ- stem, spikelet and developing grain of plants grown under control ences between treatment and the respective control conditions (low (HN-WW), low N (LN) and drought (D) conditions collected at dif- N/high N; well-watered/drought) as determined by Student’s t-test ferent time points (see Fig. S1a for details). Transcript abundance of using the Two-stage linear step-up procedure of Benjamini, Krieger the genes of interest were determined by quantitative real-time PCR and Yekutieli, with Q = 5% within a given time point. Red indicates (qRT-PCR). Calibrated normalised relative quantity (CNRQ) was cal- up-regulation and blue down-regulation with respect to the control. culated using the most stable reference genes across tissues: TaActin Time points highlighted in green indicate that significant changes in and TaGAPdH for YFEL and spikelet; TaCyclophilin and TaGAPdH allantoin concentration were detected between treatment and control for flag leaf and stem;TaCyclophilin and TaEFA2 for developing conditions according to two-way ANOVA with Tukey’s test (p < 0.05)

1 3 Plant Molecular Biology (2019) 99:477–497 487

Transcription of purine catabolic genes under low N the tissue analysed (e.g. were down-regulated in stem and and drought stress up-regulated in the spikelet). Analysis of the transcriptional profile of purine cata- To assess stress responsiveness of wheat purine catabolic bolic genes in the YFEL of Mace are largely in agreement genes, transcript abundance was quantified in all of the with those in RAC875. N deficiency increased transcrip- RAC875 samples collected during the course of the experi- tion of TaXDH1, TaALN and TaAAH in Mace YFEL; whilst ment (Fig. 5) and on YFEL (Z22) Mace samples (Fig. S4). TaALN was down-regulated andTaXDH1 up-regulated under The data for each gene and treatment is presented in Fig. 5 drought (Fig. S4). as log2 ratio of the calibrated normalised relative quantities (CNRQ) (Hellemans et al. 2007) between treatment (LN- Metabolite signature of spikelet and developing WW or HN-D) and control conditions (HN-WW). For this grain under drought and low N conditions analysis qRT-PCR primers were designed to amplify all three homeologs of each purine catabolic gene. The only To assess the involvement of allantoin in N and C metabo- exception was the design of primers to specifically amplify lism in wheat under drought and N limitation during grain TaXDH2-6DS, which is the only TaXDH2 homeolog that filling, the metabolic signatures from the spikelet and the is expressed (Fig. S3a). This likely explains why the abun- developing grain of RAC875 and Mace were assessed and dance of TaXDH2-6DS transcripts in RAC875 was approxi- visualised by hierarchical clustering (Figs. 6, 7) and PCA mately 20 times lower than in TaXDH1 (Fig. S3b). (Supplementary Fig. S5 and S6). Analysis of the transcriptional regulation of purine catab- This analysis revealed that allantoin accumulation under olism under LN-WW revealed that purine catabolic genes drought conditions (HN-D) at day 8 showed the same pat- were up-regulated in RAC875 in most tissues and devel- tern with additional metabolites forming a distinct cluster in opmental stages analysed (Fig. 5). In particular, the genes both the spikelet (D.S.ii; Fig. 6a) and the developing grain TaALN and TaAAH, putatively coding for the key enzymes (D.Dg.ii; Fig. 6b) of RAC875 and Mace. In both tissues, involved in allantoin and allantoate degradation, respec- these clusters contained mostly amino acids including the tively, were highly responsive to the LN-WW treatment. The well-known drought responsive amino acids proline and differential expression of purine catabolic genes in RAC875 4-amino-butanoic acid (GABA) and, in the developing grain, flag leaves, stem and spikelets at 17DAA corresponded to the polyamine putrescine. Generally, metabolites belonging a significant reduction of allantoin measured in the same to these clusters appeared to accumulate with higher mag- tissues as presented in Fig. 4a (time point highlighted in nitude in RAC875 compared to Mace. A genotype-specific green in Fig. 5). At 19DAA, all purine catabolic genes were response was also apparent in two additional clusters in significantly up-regulated in flag leaves and, with a lower developing grain with metabolites specifically accumulated magnitude, also in stems and spikelets (Fig. 5). (D.Dg.i) and reduced (D.Dg.iv) in RAC875 under drought, In contrast to the transcriptional response under low N respectively (Fig. 6b). conditions, RAC875 droughted plants displayed a differ- Clusters containing metabolites with an apparent reduc- ential regulation of specific sets of purine catabolic genes tion under drought in both genotypes were also identified, (Fig. 5). Generally, TaALN was down-regulated in all ana- in both the spikelet (D.S.i; Fig. 6a) and the developing lysed tissues of droughted plants (HN-D) relative to plants grain (D.Dg.iii; Fig. 6b).These clusters contained sucrose, grown under HN-WW, except in the flag leaves and spikelets myo-inositol and quinic acid. In addition, in spikelets, this where no significant changes in transcription of TaALN were cluster contained other organic acids (e.g., isocitric acid, detected. A decrease in transcript abundance of TaALN in 3-phosphate-glyceric acid) as well as the tri-saccharides raf- the YFEL occurred already after exposure to mild drought finose and 1-kestose. The cluster in developing grain further conditions (day 3), suggesting that this gene is particularly contained glucose-6-phosphate and fructose-6-phosphate, as drought responsive in this tissue (Fig. 5). TaALN down-reg- well as tryptophan and glutamate. ulation at day 6 in YFEL corresponded with accumulation Analysis of the metabolite signature under low N condi- of allantoin as presented in Fig. 4b (time point highlighted tion revealed contrasting patterns compared to the drought in green in Fig. 5). The TaXDH1 and TaUOX genes, coding treatment. Generally, the drought treatment at day 8 clearly for the enzymes that synthesise allantoin, were up-regulated separated metabolites between the two treatments (HN- under HN-D across the analysed tissues, except in devel- WW and HN-D; Fig. S5). However, the low N treatment oping grain (Fig. 5). Interestingly, TaXDH2 showed the appeared to influence the metabolite levels to a lower opposite transcriptional regulation to its paralog, TaXDH1. extent as supported by PCA analysis (Fig. S6). In par- In fact, TaXDH2 was down-regulated under HN-D in the ticular, in spikelet the major dividing component of the YFEL, the flag leaf and the stem (Fig. 5). TaAAH and TaU- data points was PC2 that separated according to the geno- GAH presented different drought responses depending on type (Fig. S6a). On the other hand, PCA of developing

1 3 488 Plant Molecular Biology (2019) 99:477–497

(A)

(B)

1 3 Plant Molecular Biology (2019) 99:477–497 489

◂Fig. 6 Metabolite response of RAC875 and Mace spikelet and Allantoin accumulation in the grain developing grain at day 8 of the drought treatment. Metabolite lev- els measured with GC-MS, ion chromatography and HPLC were log transformed and scaled by subtracting the median metabolite value To assess the contribution of allantoin to the overall N pools in each metabolite distribution. Hierarchical clustering and heatmap in the grain under control (HN-WW) and stress conditions of metabolite levels of RAC875 and Mace spikelet a and developing (HN-D and LN-WW), the plants minus the sampled tiller grain b at day 8 of the drought treatment was performed with the sup- (Supplementary Fig. S1) were grown until maturity, cor- port of ClustVis web tool (Metsalu and Vilo 2015). Pearson’s correla- tion distance of scaled data was used for the hierarchical clustering. responding overall to 16–18 plants per treatment. Total N, Each column represents a biological replicate. Representative cluster allantoin and free amino acid concentrations were then deter- containing allantoin (measured with GC-QqQ method) is noted with mined in grain harvested from the tallest remaining tiller a green line, other representative clusters are noted with a blue line (Fig. 8). Analysis of grain N% highlighted significant differ- ences between treatments (two-way ANOVA, p < 0.0001), in grain revealed no major differences between genotypes particular, it showed a 20% reduction in grain N% of plants and treatment, although a weak separation according to grown under LN-WW with no evident genotypic differences the time point, especially for 22 DAA, was detected along between RAC875 and Mace (Fig. 8a). Allantoin concentra- PC1 (Fig. S6b). tion was significantly responsive to the applied treatments Hierarchical clustering of the spikelet data under dif- and genotypes (two-way ANOVA, Genotype × Treatment, ferent N conditions placed allantoin into a representative p = 0.0028) (Fig. 8b). Particularly, allantoin was reduced cluster (Ln.S.i) which contained a large number of metab- under LN-WW by 44% and 49% in RAC875 and Mace olites, mostly carbon-rich compounds (Fig. 7a). Specifi- plants, respectively. In contrast, under HN-D, allantoin in cally, this allantoin cluster contained eleven organic acids RAC875 grains increased by 39% relatively to HN-WW. (including citric acid, malic acid, succinic acid, glyceric Interestingly, the largest difference in allantoin concentration acid and shikimic acid) and ten sugars (including fructose, between genotypes was observed under HN-WW, as previ- sucrose, myo-inositol, raffinose, 1-kestose). These metabo- ously observed for the flag leaf (Fig. 4). In fact, Mace grains lites appeared to accumulate to higher levels in spikelet of accumulated 66% more allantoin than RAC875, whilst there Mace as compared to RAC875, which showed a small or were no significant differences between the genotypes under no treatment effect (Fig. 7a). LN-WW and HN-D, as mentioned above (Fig. 8b). Similarly, in the developing grain allantoin grouped in Analysis of the free amino acids content of grain from a representative cluster (Ln.Dg.iii) which was composed plants grown under HN-WW revealed that allantoin accu- of allantoin and three carbon-rich metabolites, including mulated to levels comparable to other amino acids, such raffinose, galactinol and 1-kestose. This is in contrast to as glutamine, aspartate, asparagine and arginine (Fig. 8c). the metabolite profile in the drought treatment, where When considering that each allantoin molecule contains allantoin grouped with amino acids. These amino acids, four N atoms (1:1 nitrogen to carbon ratio, N:C), the over- under low N conditions, are scattered between two repre- all N stored in allantoin in grain was even larger than the sentative clusters in spikelets (Ln.S.ii and Ln.S.iii; Fig. 7a) N present in glutamate and aspartate, which were the most or, in developing grain, largely not assigned to any repre- abundant free amino acids identified but have only one N sentative cluster (Fig. 7b). Although, in the developing atom (1:4 N:C) (Fig. 8c). However, the N-rich amino acid grain, this large area clustered according to the time points arginine (2:3 N:C) retained approximately double the N (DAA). content of allantoin in both RAC875 and Mace grains. The However, the accumulation of allantoin, galactinol and analysis also showed significant genotypic differences in cer- raffinose in developing grain under N starvation appears tain amino acids between RAC875 and Mace. In particular, to be developmental rather than representing a specific Mace had 70% and 31% higher concentration of arginine response to N starvation as similar metabolic changes and alanine than RAC875, whilst RAC875 had 179%, 43% occurred also under control conditions at 22DAA (Fig. 8b). and 35% higher concentration of tryptophan, aspartate and In agreement with the quantitative data on allantoin shown serine than Mace, respectively (Fig. 8c). in Fig. 4a, Mace appeared to accumulate higher levels of allantoin and the other metabolites in cluster Ln.Dg.iii sug- gesting genotypic differences within wheat. Discussion In summary, the metabolite profiling showed that, under drought, allantoin clustered predominantly with N-rich com- The aim of this study was to assess the role of the purine pounds (amino acids), whereas under N-starvation allan- intermediate allantoin and the purine catabolic genes toin clustered with C-rich compounds (sugars and organic under water and nutrient stress in bread wheat. For this, we acids) and therefore allantoin represented the only N-rich selected two Australian genotypes, specifically RAC875, compound. a breeding line that has been characterised as tolerant to

1 3 490 Plant Molecular Biology (2019) 99:477–497

(A)

(B)

1 3 Plant Molecular Biology (2019) 99:477–497 491

◂Fig. 7 Metabolite response of RAC875 and Mace spikelet and devel- Wheat highly accumulates allantoin under drought oping grain under High N and Low N treatments during grain filling. stress Metabolite levels measured with GC–MS, ion chromatography and HPLC were log transformed and scaled by subtracting the median metabolite value in each metabolite distribution. Hierarchical cluster- There was a significant accumulation of allantoin in all ing and heatmap of metabolite levels of RAC875 and Mace spikelet analysed tissues and genotypes of wheat plants exposed to a and developing grain b at 17, 19 and 22 days after anthesis (DAA) drought (Fig. 4b). This was accompanied by a significant was performed with the support of ClustVis web tool (Metsalu and TaALN Vilo 2015). Pearson’s correlation distance of scaled data was used reduction in transcript levels of , i.e., the first step for the hierarchical clustering. Dashed lines represent different time in allantoin degradation (Fig. 5). points within a given genotype and treatment. Each column repre- Allantoin was previously reported to accumulate in sents a biological replicate. Representative cluster containing allan- wheat and rice in response to abiotic stresses, includ- toin (measured with GC-QqQ method) is noted with a green line, other representative clusters are noted with a blue line ing drought (Bowne et al. 2011; Degenkolbe et al. 2013; Casartelli et al. 2018), however, these studies did not investigate the regulation of the purine catabolic genes drought (Izanloo et al. 2008; Bennett et al. 2012; Bonneau under those conditions. et al. 2013) and the locally adapted variety Mace, which Analysis of the expression of purine catabolic genes in is suggested to have a higher N-use efficiency based on Arabidopsis (Irani and Todd 2016) recently showed that leaf preliminary studies (Mahjourimajd et al. 2016). accumulation of allantoin upon drought stress was associated with transcriptional up-regulation of the purine catabolic genes leading to allantoin synthesis (AtXDH1, AtXDH2, The wheat purine catabolic genes show high AtUOX, AtAS) whilst expression of the allantoin-degrading synteny but also differences compared with other gene AtALN was only marginally increased. Similarly, Yes- grasses bergenova et al. (2005) showed that, in tomato, LeXDH1 and LeXDH2 were up-regulated under drought in leaf and The genes of the purine catabolic pathway, identified in root tissues. In Arabidopsis, the allantoin pathway was also several grass genomes including hexaploid bread wheat, implicated with salt stress, showing reduced AtALN and showed a high degree of synteny among three grass AtAAH expression and increased AtXDH1 expression (Irani genomes, suggesting that the identified wheat loci are and Todd 2016; Hesberg et al. 2004), as well as an accu- true gene orthologues (Fig. 2). However, the poor synteny mulation of allantoin accompanied by AtUOX and AtALNS displayed by XDH and its adjacent genes even in diploid up-regulation and AtALN down-regulatoin (Lescano et al. genomes (Brachypodium, rice and sorghum) suggests that 2016). XDH is located in an unstable genomic region prone to Our data agree with the above studies confirming that rearrangements. The majority of the grass genomes had allantoin significantly accumulates under drought (Fig. 4b) only one copy of the purine catabolic genes (Table S5), and that this is paralleled by the up-regulation of genes puta- corresponding to three homeologs in wheat (Table S1), tively encoding enzymes for allantoin synthesis (TaXDH1 with the only exceptions of the fore-mentioned XDH, and and TaUOX) and/or down-regulation of TaALN (Fig. 5). UAH, for which there was no synteny for the Brachypo- Although we have not quantified xanthine, the gene expres- dium orthologs BdUAH2. sion data suggest that under drought, an increased amount For XDH we have identified a second copy on chromo- of xanthine is likely to feed into this pathway which, in com- some 6 of which, based on in-silico expression analysis bination with a reduced allantoin degradation, results in the (Fig. S3a), only the homeolog on chromosome 6, TaXDH2- accumulation of allantoin under drought which was observed 6DS is a functional gene explaining its low expression in our study. level compared with TaXDH1 (Fig. S3b; Fig. 5). Inter- In contrast, studies in common bean (Phaseolus vul- estingly, TaXDH2 shows an opposite transcriptional garis) showed up-regulation of both allantoin-synthetising response to drought (reduced transcript level) compared genes (PvUOX) and allantoin-degrading genes (PvALN and with TaXDH1 suggesting different roles and/or functional PvAAH) in leaves under drought and this corresponded to divergence. A differential response to drought and other increased levels of allantoin and allantoate (Alamillo et al. treatments (salt, cold, ABA) has also been shown in Arabi- 2010; Coleto et al. 2014). Interestingly, Coleto et al. (2014) dopsis, which carries a tandem duplication of AtXDH1 and reported that these ureides were more concentrated in the AtXDH2 on a single chromosome (Hesberg et al. 2004). tissues of drought-sensitive genotypes. This is in contrast to It will therefore be interesting to investigate the role and our data and other studies showing that allantoin specifically enzymatic function of TaXDH2 in more detail in relation accumulates in drought tolerant genotypes of wheat, rice and to drought and other stresses. resurrection plants (Bowne et al. 2011; Oliver et al. 2011; Degenkolbe et al. 2013; Yobi et al. 2013; Casartelli et al.

1 3 492 Plant Molecular Biology (2019) 99:477–497

Fig. 8 Total N and N-containing metabolites in mature grain of bread RAC875 and Mace grown under control conditions (HN-WW). wheat genotypes RAC875 and Mace. Comparison of a N concen- Amino acid concentration is reported in black, whilst N atoms present tration (%) and b allantoin concentration (nmol ­g−1 DW) in mature in each amino acid in grey, values are expressed as nmol ­g−1 DW. grain harvested from the tallest remaining tiller of RAC875 and Mace Allantoin and corresponding N concentration are reported in lighter plants grown under control conditions (HN-WW), drought (HN-D) colours to allow comparison. Asterisks denote significant differences and low N (LN-WW) conditions. Letters indicate significant differ- between RAC875 (R) and Mace (M) by Student’s t-test (*p < 0.05; ences between genotypes and treatments by two-way ANOVA with **p < 0.01; ****p < 0.0001). All data are mean ± SEM of 16–18 bio- Tukey’s test (p < 0.05); c free amino acids levels in mature grain of logical replicates

2018). This suggests that purine catabolism is regulated dif- supplementation, stimulated ABA and jasmonic acid (JA) ferently under drought stress in ureidic legumes. metabolism, key components of abiotic stress responses. Evidence of the functional role of purine catabolism However, a previous study suggested that allantoin does not under abiotic stress were revealed using reverse genet- possess any in-vitro antioxidant activity (Wang et al. 2012), ics approaches. Arabidopsis XDH mutants (xdh) showed prompting to speculate that allantoin and purine catabolism that disrupting the first step of purine catabolism caused play a role in stress sensing and regulation rather than a hypersensitivity to water stress (Watanabe et al. 2010) and direct ROS scavenging function. impaired recovery from extended dark exposure (Brychkova Metabolomic profiling of the spikelet (defined here as et al. 2008), which in both cases led to excessive reactive vegetative part of the spike) and the developing grain after oxygen species (ROS) accumulation compared to WT plants. 8 days of drought stress showed that allantoin clustered with In contrast, the constitutive accumulation of the intermedi- a set of highly drought-responsive metabolites (Fig. 6a, b). ate allantoin in Arabidopsis ALN knock-out mutants (aln) These clusters were composed mainly of amino acids, with led to enhanced performance under dehydration, drought alanine, valine, leucine, and proline common to the allantoin and salt stress, which also corresponded to reduced ROS clusters in spike samples (D.S.ii) and developing grain (D. levels (Watanabe et al. 2014; Irani and Todd 2016). These Dg.ii). Proline is amongst the best characterized drought- studies further showed that increasing allantoin concentra- related amino acids and has been shown to have a range tion, either constitutively (aln mutants) or by exogenous of functions under stress, e.g., osmolyte, regulator of redox

1 3 Plant Molecular Biology (2019) 99:477–497 493 potential, molecular chaperone, ROS scavenger and signal- purines ling molecule (Yoshiba et al. 1995; Hare and Cress 1997; Verbruggen and Hermans 2008; Szabados and Savoure xanthine 2010; Mohanty and Matysik 2001; Khedr et al. 2003). The XDH specific accumulation of allantoin under drought in a wide UOX range of different plant species and the suggested role in ALNS stimulating the ABA and JA pathway (see above) justifies further investigations into the importance of allantoin in the allantoin mitigation or tolerance to drought. N limitaon ALN Drought Preventing N losses under drought and allantoin AAH glutamine as a source of N under nutrient deprivation UGAH GS-GOGAT GS-GOGAT UAH Under drought, RNA and DNA as well as protein degrada- + + tion and nutrient remobilisation caused by premature leaf 4NH4 4NH4 glyoxylate senescence (Munné-Bosch and Alegre 2004) and increased Toxic build-up photorespiration (Mattsson et al. 1997; Wingler et al. 1999; NH volalizaon + 3 Kumagai et al. 2011b) are sources of high tissue NH­ 4 and related emission of volatile ammonia (Mattsson and Sch- Fig. 9 joerring 2002). In principle, free NH­ + can be recycled by Schematic model for the dual-role of allantoin under stress in 4 relation to N homeostasis. Under low N conditions, increased expres- the GS-GOGAT cycle (Fig. S2), however, this pathway is + sion of ureide-degrading genes ALN and AAH indicate that NH­ 4 is negatively responsive to drought (Nagy et al. 2013; Singh liberated from purines providing an internal source of organic N that and Gosh 2013) and might therefore not be sufficiently effec- can be re-assimilated by the GS-GOGAT cycle. As a result, allantoin tive in capturing NH­ +. Increased ammonia emission due to concentration in plant tissues is reduced. In contrast, during drought 4 stress, allantoin accumulates due to down-regulation of the ALN gene inhibition of GS with MSO has been demonstrated (Matts- + preventing the liberation of ­NH4 that could lead to cellular toxic- son and Schjoerring 1996) and has been directly linked ity and part of it might be lost as volatile ammonia due to drought- with enhanced photorespiration under high light and high impaired GS-GOGAT activity ­O2 stress (Kumagai et al. 2011b), as well as with senescence (Parton et al. 1988) and heat stress (Mattson et al. 1997). Ammonia emission can be considered an efficient, though drought-tolerant specific GS responses have also been shown very wasteful mechanism, to prevent the build-up of high tis- in tomato (Sanchez-Rodriguez et al. 2011) and wheat (Nagy sue concentrations, which is toxic to plants (for a review see et al. 2013).We suggest that the selection of high allantoin Britto and Kronzucker 2002). The accumulation of allantoin and maintenance of GS activity (reduced N losses) under (e.g. 30-fold accumulation in RAC875 flag leaves, Fig. 4b) drought stress could be a specific target for breeding (Fig. 9). would therefore be beneficial to plants because it prevents The importance of allantoin and purine catabolism in N + accumulation of ­NH4 to toxic levels and also, to some metabolism have long been established in ureidic legumes, extent, to retain organic N in the plant that could be lost to which employ allantoin and allantoate as major forms for the atmosphere in the form of volatile ammonia. Although long-distance transport of N (Schubert 1986; Sinclair and allantoin represents just a fraction of the total organic N in a Serraj 1995). However, recently an important role in main- wheat plant, allantoin accumulation under drought appears taning N homeostasis in non-ureidic plant species is becom- to occur simultaneously with the accumulation of several ing evident. Soltabayeva et al. (2018) reported that Arabi- other small metabolites containing N, notably amino acids dopsis Atxdh1, Ataln and Ataah mutants displayed an early − such as proline (Fig. 6). This response seems to be shared senescence phenotype when grown under low NO­ 3 con- by other plant species (see Introduction for references), in ditions and that the activity of nitrate reductase (NR) was fact we previously reported high accumulation of allan- increased in leaves of Atxdh1 mutants suggesting a higher toin and several amino acids when rice plants were sub- demand for nitrate than wild type plants. Similarly, Naka- jected to drought stress (Casartelli et al. 2018). Therefore, gawa et al. (2007) reported an early onset of senescence in this could underlie a global strategy that plants adopt to RNAi-induced xdh mutants which also displayed reduced improve N balance under stress, when GS-GOGAT enzy- chlorophyll content and increased cytosolic GS1 protein, matic activity is reduced. Differences in the regulation of GS which is known to be induced during senescence (Bernard (Singh and Gosh 2013) and ammonia emission (Kumagai et al. 2008). et al. 2011) between drought tolerant and intolerant gen- In our study, the quantification of allantoin in the otypes have been shown in rice and genetic diversity and wheat plants grown under N-limited conditions revealed

1 3 494 Plant Molecular Biology (2019) 99:477–497 a significant reduction of allantoin in all tissues analysed the majority of metabolites that showed an opposite trend (Fig. 4a) which is in contrast to the observed accumulation (Fig. 7). of allantoin under drought (Fig. 4b) as discussed above. A Although we calculated that allantoin accounted for only decrease of allantoin under N deprivation was also reported about 0.3% of the total N measured in wheat grains, this in other cereal species, specifically in maize and rice show- represents a significant portion of the soluble N pool, com- ing reduced allantoin in leaves and roots, respectively (Ami- parable to the most concentrated and N-rich amino acids, our et al. 2012; Coneva et al. 2014). In addition, high induc- such as asparagine and arginine (Fig. 8c). Previous studies tion of OsALN expression under low N conditions in rice have shown that allantoin stored in wheat grains is quickly was recently shown (Lee et al. 2018), this was previously utilised, from as early as 1 day after germination (Montalbini also observed in Arabidopsis and the leguminous tree Rob- 1992), suggesting that it can be used as a readily available inia pseudoacacia (Yang and Han 2004). In agreement with N substrate. this, the reduction of allantoin under low N reported in this The data reported here further show that genotypic dif- study (Fig. 4a) was accompanied by increased transcription ferences exist among bread wheat genotypes, in fact, Mace of purine catabolic genes, and this was particularly evident accumulated 66% more grain allantoin than RAC875, in genes encoding for the ureide degrading enzymes TaALN despite similar % grain N (Fig. 8a, b). This difference was and TaAAH (Fig. 5). also particularely evident in flag leaves of Mace, which had Downstream compounds of the purine catabolic path- up to 10-fold higher allantoin content compared to RAC875 way were recently analysed in more detail under N depriva- under both, control and low N conditions (Fig. 4a). Mace tion using the same wheat genotypes, RAC875 and Mace is widely grown in Australia because of its superior yield (Melino et al. 2018) showing, in addition to reduced levels of under drought and nitrogen-use efficiency (Mahjourimajd allantoin, a significant reduction of allantoate in leaves and et al. 2016). High allantoin concentrations in the leaves an accumulation of glyoxylic acid, the end product of purine and grain might indeed be a positive contributing factor to degradation. The assumption that the N (and C) remobilised high NUE in Mace. This is in agreement with findings in from allantoin supports plant growth is supported by studies rice where differences in allantoin levels in genotypes with showing that Arabidopsis and rice seedlings could grow with contrasting drought tolerance were already apparent under ureides as a sole N source (Desimone et al. 2002; Brychkova control conditions (Casartelli et al. 2018). In addition to et al. 2008; Lee et al. 2018), although growth was delayed greater accumulation of allantoin, mature grains of Mace in comparison to plants supplied with inorganic N. This is also accumulated 70% more arginine than RAC875 (Fig. 8c). in contrast to wheat seedlings that, resupplied with allantoin Arginine is the proteinogenic amino acid with the highest − and xanthine as a sole source of N after NO­ 3 starvation, N:C ratio (3:2), known to represent a major N form for stor- grew and photosynthesised as well as those re-supplied with age and transport. − NO­ 3 (Melino et al. 2018). Our data thus indicate that the two analysed wheat geno- Taken together, these findings suggest that the enhanced types prioritize different N compounds for transport and activity of the purine catabolic pathway provides an internal grain loading. Previous studies in Mace suggest that this source of organic N used to maintain homeostasis under cultivar has high N use efficiency (Mahjourimajd et al. 2016) low N conditions whereas the accumulation of allantoin and this study and Kastury et al. (2018) additionally dem- under drought releases pressure from the GS-GOGAT cycle onstrated that Mace is higher yielding under drought than + thereby preventing accumulation of toxic levels of NH­ 4 and RAC875. The possibility that the superior performance of possibly N losses due to volatilization (Fig. 9). Mace may be related to the preferential use of metabolites with high N:C ratio, which are more energy effective forms Allantoin is a relevant component of grain nitrogen for transporting and storing N, is an intriguing hypothesis in wheat that requires further validation. In addition, allantoin and arginine may also play a regulatory role in plant growth and In wheat, N translocated to the grain during plant senes- development as they participate in ABA, JA, and nitric oxide cence is an important determinant of grain quality and there (NO) signalling, respectively (Watanabe et al. 2014; Winter is currently little information on the contribution of allan- et al. 2015; Takagi et al. 2016). toin and/or other purine catabolites to grain N in non-ureide plants. The data reported here show progressive accumula- tion of allantoin during grain filling (Fig. 4) up to more than Summary 1400 nmol ­g−1 DW at maturity in Mace (Fig. 8b). This was confirmed by metabolomics analyses of developing grains In summary, our data are in support of the converging showing that allantoin levels correlated with other metab- evidence that allantoin has a relevant role in non-ureide olites specifically accumulating at 22DAA, in contrast to plants under water stress and in N homeostasis. In addition

1 3 Plant Molecular Biology (2019) 99:477–497 495 to the proposed role of allantoin in drought-induced ABA References responses, accumulation of this N-rich molecule under drought would contribute to optimise N (and C) balance in Alamillo JM, Diaz-Leal JL, Sanchez-Moran MV, Pineda M (2010) the plant by preventing toxic build-up of NH­ + and possible Molecular analysis of ureide accumulation under drought stress in 4 Phaseolus vulgaris L. PlantCell. Environ 33:1828–1837 N losses through volatile ammonia due to a reduced activ- Amiour N, Imbaud S, Clément G et al (2012) The use of metabolomics ity of the GS-GOGAT cycle (Fig. 9). Under low N condi- integrated with transcriptomic and proteomic studies for identify- tions, induced expression of allantoin catabolic genes and ing key steps involved in the control of nitrogen metabolism in decreased allantoin levels indicate that allantoin serves as crops such as maize. J Exp Bot 63:5017–5033 Barbottin A, Lecomte C, Bouchard C, Jeuffroy M-H (2005) Nitro- an internal organic N source (Fig. 9). In wheat, allantoin is gen remobilization during grain filling in wheat: genotypic and also a relevant component of grain N and comparable to that environmental effects. Crop Sci 45:1141 of other major amino acids. Bennett D, Izanloo A, Reynolds M, Kuchel H, Langridge P, Schn- Sufficient evidence for genetic diversity for the purine urbusch T (2012) Genetic dissection of grain yield and physi- cal grain quality in bread wheat (Triticum aestivum L.) under catabolic pathway in relation to drought tolerance and water-limited environments. Theo Appl Genetics 125:255–271 enhanced NUE is now available and, in conjunction with Bernard SM, Møller ALB, Dionisio G et al (2008) Gene expres- reported genetic diversity for GS-GOGAT, provides an sion, cellular localisation and function of glutamine synthetase Triticum aestivum opportunity to explore this further for the development of isozymes in wheat ( L.). Plant Mol Biol 67:89. https​://doi.org/10.1007/s1110​3-008-9303-y crops with enhanced drought tolerance and NUE. Bolger AM, Lohse M, Usadel B (2014) Trimmomatic: a flexible trim- mer for Illumina sequence data. Bioinformatics 30:2114–2120 Acknowledgements We thank Akiko Enju, Pia Müller and Jessey Bonneau J, Taylor J, Parent B, Bennett D, Reynolds M, Feuillet C, George (University of Adelaide) for supporting with sample collec- Langridge P, Mather D (2013) Multi-environment analysis and tion; Juan Carlos Sanchez-Ferrero (University of Adelaide) for sup- improved mapping of a yield-related QTL on chromosome 3B port with bioinformatic analyses; Adam Lukaszewski (University of of wheat. Theo Appl Genetics 126:747–761 Riverside, California, US) for providing the Chinese Spring NT seeds Bowne JB, Erwin TA, Juttner J, Schnurbusch T, Langridge P, Bacic stock and Margaret Pallotta (University of Adelaide) for providing A, Roessner U (2011) Drought responses of leaf tissues from DNA of the NT lines; Sanjiv Satija, Larissa Chirkova and Yuan Li wheat cultivars of differing drought tolerance at the metabolite (University of Adelaide) for technical support with molecular analy- level. Mol Plant 5:418–429 + ses; Julian Taylor (University of Adelaide) for support in experimental Britto DT, Kronzucker HJ (2002) NH­ 4 toxicity in higher plants: a design; Siria Natera, Gina Barossa and Veronica Liu (Metabolomics critical review. Plant Physol 159:567–584 Australia) for support with the allantoin quantification. We are grateful Brychkova G, Fluhr R, Sagi M (2008) Formation of xanthine and the to Elmien Heyneke, Ines Fehrle and Astrid Basner (Max Plank Insti- use of purine metabolites as a nitrogen source in Arabidopsis tute, Potsdam-Golm) for their support in performing the metabolomics plants. Plant Signal Behav 3:999–1001 analysis. This work was supported by an Australian Research Coun- Casartelli A, Riewe D, Hubberten HM, Altmann T, Hoefgen R, cil linkage Grant LP1400100239 in partnership with and additional Heuer S (2018) Exploring traditional aus-type rice for metabo- funding from DuPont-Pioneer (USA). Rothamsted Research receives lites conferring drought tolerance. Rice 11:9 strategic funding from the Biotechnological and Biological Sciences Chomczynski P (1993) A reagent for the single-step simultaneous Research Council of the United Kingdom. We acknowledge support isolation of RNA, DNA and proteins from cell and tissue sam- through the Designing Future Wheat (DFW) Strategic Programme (BB/ ples. Biotechniques 15:532–534 P016855/1). Choulet F, Alberti A, Theil S, Glover N, Barbe V, Daron J, Pin- gault L, Sourdille P, Couloux A, Paux E (2014) Structural and Author Contributions SH was leading the project. MO, VJM and RH functional partitioning of bread wheat chromosome 3B. Science conceived the ideas, edited the manuscript and co-supervised A.C. 345:1249721 who conducted the experiments as part of his PhD thesis. AC wrote the Clavijo BJ, Venturini L, Schudoma C et al (2017) An improved manuscript with inputs from SH, MO, VJM and RH. The quantitative assembly and annotation of the allohexaploid wheat genome allantoin method was developed by UR, NSJ and HM. The identifica- identifies complete families of agronomic genes and provides tion of the wheat genes and bioinformatics analyses was supported by genomic evidence for chromosomal translocations. Genome Res UB and RS. Inputs on the quantitative gene expression analyses was 27:885–896 provided by MR. Access to analytical pipelines, technical support and Coleto I, Pineda M, Rodino AP, De Ron AM, Alamillo JM (2014) data analysis was provided by AE, MW, and EZ. Comparison of inhibition of ­N2 fixation and ureide accumulation under water deficit in four common bean genotypes of contrasting Open Access This article is distributed under the terms of the Crea- drought tolerance. Ann Botany 113:1071–1082 tive Commons Attribution 4.0 International License (http://creat​iveco​ Coneva V, Simopoulos C, Casaretto JA et al (2014) Metabolic and mmons.org/licen​ ses/by/4.0/​ ), which permits unrestricted use, distribu- co-expression network-based analyses associated with nitrate tion, and reproduction in any medium, provided you give appropriate response in rice. BMC Genom 15:1056 credit to the original author(s) and the source, provide a link to the Degenkolbe T, Do PT, Kopka J, Zuther E, Hincha DK, Köhl KI (2013) Creative Commons license, and indicate if changes were made. Identification of drought tolerance markers in a diverse population of rice cultivars by expression and metabolite profiling. PLoS ONE 8:e63637 Desimone M, Catoni E, Ludewig U, Hilpert M, Kunze AS, Tegeder R, Frommer M, Schumacher WB (2002) A novel superfamily of transporters for allantoin and other oxo derivatives of nitrogen heterocyclic compounds in Arabidopsis. Plant Cell 14:847–856

1 3 496 Plant Molecular Biology (2019) 99:477–497

Do PT, Degenkolbe T, Erban A, Heyer AG, Kopka J, Köhl KI, Hincha Lee D-K, Redillas MCFR, Jung H, Choi S, Kim YS, Kim J-K (2018) DK, Zuther E (2013) Dissecting rice polyamine metabolism under A nitrogen molecular sensing system, comprised of the ALLAN- controlled long-term drought stress. PLoS ONE 8:e60325 TOINASE and UREIDE PERMEASE 1 genes, can be used to Dobin A, Davis CA, Schlesinger F, Drenkow J, Zaleski C, Jha S, Batut monitor N status in rice. Front Plant Sci 9:444 P, Chaisson M, Gingeras TR (2013) STAR: ultrafast universal Lescano C, Martini C, González C, Desimone M (2016) Allantoin RNA-seq aligner. Bioinformatics 29:15–21 accumulation mediated by allantoinase downregulation and trans- Erban A, Schauer N, Fernie AR, Kopka J (2007) Nonsupervised con- port by Ureide Permease 5 confers salt stress tolerance to Arabi- struction and application of mass spectral and retention time index dopsis plants. Plant Mol Biol 91:581–595 libraries from time-of-flight gas chromatography–mass spectrom- Li H, Handsaker B, Wysoker A, Fennell T, Ruan J, Homer N, Marth G, etry metabolite profiles. Humana Press, New York pp 19–38 Abecasis G, Durbin R (2009) The sequence alignment/map format Gill BS, Appels R, Botha-Oberholster A-M et al (2004) A work- and SAMtools. Bioinformatics 25:2078–2079 shop report on wheat genome sequencing: international genome Lisec J, Schauer N, Kopka J, Willmitzer L, Fernie AR (2006) Gas research on wheat consortium. Genetics 168:1087–1096 chromatography mass spectrometry-based metabolite profiling Hanks JF, Tolbert N, Schubert KR (1981) Localization of enzymes of in plants. Nat Protoc 1:387 ureide biosynthesis in peroxisomes and microsomes of nodules. Mahjourimajd S, Kuchel H, Langridge P, Okamoto M (2016) Evalu- Plant Physiol 68:65–69 ation of Australian wheat genotypes for response to variable Hare P, Cress W (1997) Metabolic implications of stress-induced nitrogen application. Plant Soil 399:247–255 proline accumulation in plants. Plant Growth Regul 21:79–102 Masclaux-Daubresse C, Daniel-Vedele F, Dechorgnat J, Chardon F, Hellemans J, Mortier G, De Paepe A, Speleman F, Vandesompele J Gaufichon L, Suzuki A (2010) Nitrogen uptake, assimilation (2007) qBase relative quantification framework and software for and remobilization in plants: challenges for sustainable and management and automated analysis of real-time quantitative productive agriculture. Ann Bot 105:1141–1157 PCR data. Genome Biol 8:R19 Mattsson M, Schjoerring JK (1996) Ammonia emission from young Herridge DF, Atkins CA, Pate JS, Rainbird RM (1978) Allantoin and barley plant: influence of N-source, light/dark cycles and inhibi- Allantoic Acid in the Nitrogen Economy of the Cowpea (Vigna tion of glutamine synthetase. J Exp Bot 47:477–484 unguiculata [L.] Walp.). Plant Physiol 62:495–498 Mattsson M, Schjoerring JK (2002) Dynamic and steady-state Hesberg C, Hänsch R, Mendel RR, Bittner F (2004) Tandem ori- responses of inorganic nitrogen pools and NH3 exchange in entation of duplicated xanthine dehydrogenase genes from leaves of Lolium perenne and Bromus erectus to changes in root Arabidopsis thaliana: differential gene expression and enzyme nitrogen supply. Plant Physiol 128:742–750 activities. J Biol Chem 279:13547–13554 Mattsson M, Häusler RE, Leegood RC, Lea PJ, Schjoerring JK Irani S, Todd CD (2016) Ureide metabolism under abiotic stress in (1997) Leaf-atmosphere NHâ exchange in barley mutants Arabidopsis thaliana. J Plant Physiol 199:87–95 with reduced activities of glutamine synthetase. Plant Physiol Izanloo A, Condon AG, Langridge P, Tester M, Schnurbusch T 114:1307–1312 (2008) Different mechanisms of adaptation to cyclic water Melino VJ, Fiene G, Enju A, Cai J, Buchner P, Heuer S (2015) stress in two South Australian bread wheat cultivars. J Exp Bot Genetic diversity for root plasticity and nitrogen uptake in wheat 59:3327–3346 seedlings. Funct Plant Biol 42:942 Kanani H, Dutta B, Klapa MI (2010) Individual vs. combinatorial Melino VJ, Casartelli A, George J, Rupasinghe T, Roessner U, effect of elevatedCO ­ 2 conditions and salinity stress on Arabi- Okamoto M, Heuer S (2018) RNA catabolites contribute to dopsis thaliana liquid cultures: comparing the early molecular the nitrogen pool and support growth recovery of wheat. Front response using time-series transcriptomic and metabolomic analy- Plant Sci 9:1539 ses. BMC Syst Biol 4:177–177 Metsalu T, Vilo J (2015) ClustVis: a web tool for visualizing cluster- Kaplan F, Kopka J, Haskell DW, Zhao W, Schiller KC, Gatzke N, Sung ing of multivariate data using principal component analysis and DY, Guy CL (2004) Exploring the temperature-stress metabolome heatmap. Nucleic Acids Res 43:W566–W570 of Arabidopsis. Plant Physiol 136:4159–4168 Mohanty P, Matysik J (2001) Effect of proline on the production of Kastury F, Rahimi Eichi V, Enju A, Okamoto M, Heuer S, Melino V singlet oxygen. Amino Acids 21:195–200 (2018) Exploring the potential for top-dressing bread wheat with Montalbini P (1992) Ureides and enzymes of ureide synthesis in ammonium chloride to minimize grain yield losses under drought. wheat seeds and leaves and effect of allopurinol onPuccinia Soil Sci Plant Nutr. https://doi.org/10.1080/00380​ 768.2018.14933​ ​ recondita f. sp. tritici infection. Plant Sci 87:225–231 41 Moschen S, Bengoa Luoni S, Di Rienzo JA, Caro MdP, Tohge T, Khedr AHA, Abbas MA, Wahid AAA, Quick WP, Abogadallah GM Watanabe M, Hollmann J, González S, Rivarola M, García- (2003) Proline induces the expression of salt-stress-responsive García F (2016) Integrating transcriptomic and metabolomic proteins and may improve the adaptation of Pancratium mariti- analysis to understand natural leaf senescence in sunflower. mum L. to salt-stress. J Exp Bot 54:2553–2562 Plant Biotechnol J 14:719–734 Kim K, Park J, Rhee S (2007) Structural and Functional basis for Munné-Bosch S, Alegre L (2004) Die and let live: leaf senescence (S)-allantoin formation in the ureide pathway. J Biol Chem contributes to plant survival under drought stress. Funct Plant 282:23457–23464 Biol 31:203–216 Kumagai E, Araki T, Hamaoka N, Ueno O (2011) Ammonia emis- Nagy Z, Németh E, Guóth A, Bona L, Wodala B, Pécsváradi A sion from rice leaves in relation to photorespiration and geno- (2013) Metabolic indicators of drought stress tolerance in typic differences in glutamine synthetase activity. Annals Bot wheat: glutamine synthetase isoenzymes and rubisco. Plant 108:1381–1386 Physiol Biochem 67:48–54 Lamberto I, Percudani R, Gatti R, Folli C, Petrucco S (2010) Con- Nakagawa A, Sakamoto S, Takahashi M, Morikawa H, Sakamoto served alternative splicing of arabidopsis transthyretin-like A (2007) The RNAi-mediated silencing of xanthine dehydro- determines protein localization and S-allantoin synthesis in per- genase impairs growth and fertility and accelerates leaf senes- oxisomes. Plant Cell 22:1564–1574 cence in transgenic Arabidopsis plants. Plant Cell Physiol Lea PJ, Miflin BJ (2010) Nitrogen assimilation and its relevance to crop 48:1484–1495 improvement. Ann Plant Rev 42:1–40 Nam M, Bang E, Kwon T, Kim Y, Kim E, Cho K, Park W, Kim B, Yoon I (2015) Metabolite profiling of diverse rice germplasm

1 3 Plant Molecular Biology (2019) 99:477–497 497

and identification of conserved metabolic markers of rice roots Triplett EW, Blevins DG, Randall DD (1982) Purification and proper- in response to long-term mild salinity stress. Int J Mol Sci ties of soybean nodule xanthine dehydrogenase. Arch Biochem 16:21959–21974 Biophys 219:39–46 Nikiforova VJ, Kopka J, Tolstikov V, Fiehn O, Hopkins L, Hawkesford Verbruggen N, Hermans C (2008) Proline accumulation in plants: a MJ, Hesse H, Hoefgen R (2005) Systems rebalancing of metabo- review. Amino Acids 35:753–759 lism in response to sulfur deprivation, as revealed by metabolome Wang P, Kong CH, Sun B, Xu XH (2012) Distribution and Function of analysis of Arabidopsis plants. Plant Physiol 138:304 Allantoin (5-Ureidohydantoin) in Rice Grains. J Agric Food Chem Oliver MJ, Guo L, Alexander DC, Ryals JA, Wone BW, Cushman JC 60:2793–2798. https​://doi.org/10.1021/jf205​1043 (2011) A sister group contrast using untargeted global metabo- Wang W-S, Zhao X-Q, Li M, Huang L-Y, Xu J-L, Zhang F, Cui Y-R, Fu lomic analysis delineates the biochemical regulation underly- B-Y, Li Z-K (2016) Complex molecular mechanisms underlying ing desiccation tolerance in Sporobolus stapfianus. Plant Cell seedling salt tolerance in rice revealed by comparative transcrip- 23:1231–1248 tome and metabolomic profiling. J Exp Bot 67:405–419 Parton WJ, Morgan JA, Altenhofen JM, Harper LA (1988) Ammo- Watanabe S, Nakagawa A, Izumi S, Shimada H, Sakamoto A (2010) nia volatilization from spring wheat. Plants Agron J 80:419–425. RNA interference-mediated suppression of xanthine dehydroge- https​://doi.org/10.2134/agron​j1988​.00021​96200​80000​30008​x nase reveals the role of purine metabolism in drought tolerance Pate JS, Atkins CA, White ST, Rainbird RM, Woo KC (1980) Nitro- in Arabidopsis. FEBS Lett 584:1181–1186 gen nutrition and xylem transport of nitrogen in ureide-producing Watanabe S, Matsumoto M, Hakomori Y, Takagi H, Shimada H, Saka- grain legumes. Plant Physiol 65:961–965 moto A (2014) The purine metabolite allantoin enhances abiotic Pessoa J, Sárkány Z, Ferreira-da-Silva F, Martins S, Almeida MR, Li stress tolerance through synergistic activation of abscisic acid J, Damas AM (2010) Functional characterization of Arabidopsis metabolism. Plant Cell Environ 37:1022–1036 thaliana transthyretin-like protein. BMC Plant Biol 10:30 Werner AK, Witte CP (2011) The biochemistry of nitrogen mobiliza- Ramazzina I, Folli C, Secchi A, Berni R, Percudani R (2006) Com- tion: purine ring catabolism. Trends Plant Sci 16:381–387 pleting the uric acid degradation pathway through phylogenetic Werner AK, Sparkes IA, Romeis T, Witte CP (2008) Identification, comparison of whole genomes. Nat Cheml Biol 2:144–148 biochemical characterization, and subcellular localization of Sánchez-Rodríguez E et al (2011) Ammonia production and assimila- allantoate amidohydrolases from Arabidopsis and soybean. Plant tion: Its importance as a tolerance mechanism during moderate Physiol 146:418–430 water deficit in tomato plants. J Plant Physiol 168:816–823. https​ Werner AK, Romeis T, Witte C-P (2010) Ureide catabolism in Arabi- ://doi.org/10.1016/j.jplph​.2010.11.018 dopsis thaliana and Escherichia coli. Nat Chem Biol 6:19–21 Schubert KR (1986) Products of biological nitrogen fixation in higher Wingler A, Quick WP, Bungard RA, Bailey KJ, Lea PJ, Leegood RC plants: synthesis, transport, and metabolism. Ann Rev Plant (1999) The role of photorespiration during drought stress: an Physiol 539–574 analysis utilizing barley mutants with reduced activities of pho- Sears ER (1954) The aneuploids of common wheat. Missouri Agricult torespiratory enzymes. Plant Cell Environ 22:361–373 Exp Stat Annu Bull 572:1–59 Winter G, Todd CD, Trovato M, Forlani G, Funck D (2015) Physi- Serventi F, Ramazzina I, Lamberto I, Puggioni V, Gatti R, Percudani ological implications of arginine metabolism in plants. Front Plant R (2010) Chemical basis of nitrogen recovery through the ureide Sci 6:534 pathway: formation and hydrolysis of S-ureidoglycine in plants Wu H, Liu X, You L, Zhang L, Zhou D, Feng J, Zhao J, Yu J (2012) and bacteria. ACS Chem Biol 5:203 Effects of salinity on metabolic profiles, gene expressions, and Silvente S, Sobolev AP, Lara M (2012) Metabolite adjustments in antioxidant enzymes in halophyte. J Plant Growth Reg 31:332–341 drought tolerant and sensitive soybean genotypes in response to Yang J, Han KH (2004) Functional characterization of allantoinase water stress. PLoS ONE 7:e38554 genes from Arabidopsis and a nonureide-type legume black locust. Sinclair TR, Serraj R (1995) Legume nitrogen fixation and drought. Plant Physiol 134:1039–1049 Nature 378:344–344 Yesbergenova Z, Yang G, Oron E, Soffer D, Fluhr R, Sagi M (2005) Singh K, Ghosh S (2013) Regulation of glutamine synthetase isoforms The plant Mo-hydroxylases aldehyde oxidase and xanthine dehy- in two differentially drought-tolerant rice Oryza( sativa L.) cul- drogenase have distinct reactive oxygen species signatures and are tivars under water deficit conditions. Plant Cell Rep 32:183–193 induced by drought and abscisic acid. Plant J 42:862–876 Singh G, Kumar S, Singh P (2003) A quick method to isolate RNA Yobi A, Wone BW, Xu W, Alexander DC, Guo L, Ryals JA, Oliver MJ, from wheat and other carbohydrate-rich seeds. Plant Mol Biol Cushman JC (2013) Metabolomic profiling in Selaginella lepido- Report 21:93–93 phylla at various hydration states provides new insights into the Soltabayeva A, Srivastava S, Kurmanbayeva A, Bekturova A, Fluhr mechanistic basis of desiccation tolerance. Mol Plant 6:369–385 R, Sagi M (2018) Early senescence in older leaves of low nitrate- Yoshiba Y, Kiyosue T, Katagiri T, Ueda H, Mizoguchi T, Yamaguchi- grown Atxdh1 uncovers a role for purine catabolism in N supply. Shinozaki K, Wada K, Harada Y, Shinozaki K (1995) Correlation Plant Physiol. https​://doi.org/10.1104/pp.18.00795​ between the induction of a gene for δ1-pyrroline-5-carboxylate Szabados L, Savoure A (2010) Proline: a multifunctional amino acid. synthetase and the accumulation of proline in Arabidopsis thali- Trends Plant Sci 15:89–97 ana under osmotic stress. Plant J 7:751–760 Takagi H, Ishiga Y, Watanabe S et al (2016) Allantoin, a stress-related purine metabolite, can activate jasmonate signaling in a MYC2- Publisher’s Note Springer Nature remains neutral with regard to regulated and abscisic acid-dependent manner. J Exp Bot 67:2519 jurisdictional claims in published maps and institutional affiliations. Todd CD, Polacco JC (2006) AtAAH encodes a protein with allan- toate amidohydrolase activity from Arabidopsis thaliana. Planta 223:1108–1113

1 3